CN107895376A - Based on the solar panel recognition methods for improving Canny operators and contour area threshold value - Google Patents

Based on the solar panel recognition methods for improving Canny operators and contour area threshold value Download PDF

Info

Publication number
CN107895376A
CN107895376A CN201711313921.3A CN201711313921A CN107895376A CN 107895376 A CN107895376 A CN 107895376A CN 201711313921 A CN201711313921 A CN 201711313921A CN 107895376 A CN107895376 A CN 107895376A
Authority
CN
China
Prior art keywords
image
area
solar panel
threshold
threshold value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201711313921.3A
Other languages
Chinese (zh)
Inventor
郑茜颖
周海芳
戴龙云
程树英
林锦州
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fuzhou University
Original Assignee
Fuzhou University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Fuzhou University filed Critical Fuzhou University
Priority to CN201711313921.3A priority Critical patent/CN107895376A/en
Publication of CN107895376A publication Critical patent/CN107895376A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformation in the plane of the image
    • G06T3/40—Scaling the whole image or part thereof
    • G06T3/4007—Interpolation-based scaling, e.g. bilinear interpolation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration by the use of local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration by the use of local operators
    • G06T5/30—Erosion or dilatation, e.g. thinning
    • G06T5/70—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/13—Edge detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20172—Image enhancement details
    • G06T2207/20192—Edge enhancement; Edge preservation

Abstract

The present invention relates to a kind of based on the solar panel recognition methods for improving Canny operators and contour area threshold value.First, row interpolation sampling is entered to the image of collection, reduces the size and pixel of picture；Clutter edge and noise spot, prominent target area are filtered out using gaussian filtering；Secondly, converting colors space, it is easy to extract the colouring informations such as saturation degree；Independent Point is excluded, realizes and strengthens target area, weakens the purpose of background area；Finally, using improving Canny operators acquisition dynamic threshold and carrying out rim detection to image, two-value contour images are obtained；Enter row threshold division to contour area, exclude background area, retain target area.The present invention can improve the accuracy rate of solar panel identification.

Description

Based on the solar panel identification for improving Canny operators and contour area threshold value Method
Technical field
The invention belongs to field of photovoltaic technology, and in particular to a kind of based on improving Canny operators and contour area threshold value Solar panel recognition methods.
Background technology
Solar cell is converted solar energy into electrical energy using photoelectric effect, and many battery series-parallel connections, which get up, just constitutes me Common big power output solar panel.Use in daily life becomes increasingly popular, but its complicated production Technique can cause the color of cell piece incomplete same, and the inconsistency of these outward appearances adds the difficulty of identification.Somewhere light Lie prostrate industry size and solar panel service condition, for assess regional a clean energy resource and traditional energy accounting, Photovoltaic energy service condition etc. has great importance.In order to identify solar panel, have with reference to solar panel Color characteristic, the reasonable conversion of color space is carried out, rim detection is can be carried out after exclusive PCR point and noise.
Traditional Canny operators carry out rim detection using the change of single order or Second order directional, and speed is fast, but can go out Existing details profile is lost, while meeting flase drop goes out Clutter edge, causes solar panel identification error.Improve Canny operators Emphasis is focused primarily in the selection of threshold value, and the selection of traditional Canny operator threshold values relies primarily on the experience of operating personnel, is considered To solar panel identification can by weather, background is different is influenceed, so optimal threshold is not changeless, therefore This threshold value cannot be fixed value.Enter because the identification of solar panel needs to have under different weather environment and background OK, so needing to choose suitable dynamic threshold, and then the correct detection at edge is realized.
The shape of solar panel is typically rendered as the polygon of rule in the picture, after the detection of Canny operators Image edges only edge information, the Clutter edge such as including surrounding building among these, so needing to carry out these interference informations Reject.Contours extract is a kind of Feature Extraction Technology, its extract be target area intersect with background area make the ladder to be formed Edge is spent, the edge gradient formed is higher, then profile is more clear.Due to noise and some interference edges occurring in image Edge, the appearance of these information can cause the accuracy of detection of solar panel to have a greatly reduced quality, the rule having due to solar panel Then geometry, feature extraction is carried out to it, edge contour information can be drawn out.
The extraction focused on to target area and background area boundary line of rim detection, these brightness change are compared Substantially, but due to light, the influence of background, so the threshold value set is not fixed value, and the shape of background object and If color is close with solar panel, identification error may result in.Therefore, it is necessary to which solar energy can be recognized accurately in one kind The image-recognizing method of cell panel.
The content of the invention
It is an object of the invention to provide a kind of based on the solar panel for improving Canny operators and contour area threshold value Recognition methods, dynamic threshold is determined by improving Canny operators, correctly detects the marginal information of different background, it is and then right The edge of detection carries out the segmentation of contour area threshold value, can exclude the influence of the ambient interferences such as Adjacent Buildings, correct identification Sunny energy cell panel.
To achieve the above object, the technical scheme is that：One kind is based on improvement Canny operators and contour area threshold value Solar panel recognition methods, comprise the following steps,
Step S1, the image of solar panel is gathered by image capture device, is stored in computer, and using slotting The method of value sampling is adjusted to the size and pixel of image；
Step S2, sampled via the interpolation in step S1, be filtered operation to image by the way of gaussian filtering；
Step S3, wrapped via the filtering operation in step S2, the color space of transition diagram picture in follow-up operation Include brightness, the colouring information of saturation degree；
Step S4, to step S3 converting colors space afterwards, it is necessary to be excluded to Independent Point；After exclusion Independent Point Figure carry out morphology and open operation, holding overall intensity level and larger bright features are relatively constant；
Step S5, the image obtained to step S4 processing, rim detection is carried out to image using improved Canny operators and obtained To edge image；
Step S6, the edge image obtained to step S5 processing, the wheel of its image is extracted using contour area Threshold segmentation Wide information, the contour line of solar panel is drawn out, complete the identifying purpose of solar panel.
In an embodiment of the present invention, the gaussian filtering method in the step S2 is：By the way of gaussian filtering pair The profile information of image redundancy is filtered out, and during gaussian filtering, each pixel is by other in itself and neighborhood What pixel obtained after being weighted averagely, i.e. the Gaussian Blur process of image is exactly that image does convolution with normal distribution.
In an embodiment of the present invention, in the step S5, rim detection is carried out to image using improved Canny operators The method for obtaining edge image is：
By improving the objective selected threshold of Canny operators, to cause Threshold segmentation to divide target area and background area, make The dynamic threshold that the noise spot of erroneous segmentation is as few as possible, is determined using maximum between-cluster variance must occur：
The probability shared by each gray value of image is determined first, and the probability for defining gray scale i is pi, define threshold value Th, traversal Th=0, Th=1 ... ..., Th=255；
Image is divided into by two parts according to threshold value Th, is defined as F1, F2；Wherein：F1=f (x, y) | f (x, y) >=Th }；F2 =f (x, y) | f (x, y) ＜ Th }；
F is calculated respectively1, F2The average value UTemp of respective gray scale1, UTemp2, respective shared probability P1, P2；It is defined as： UTemp1=∑ pi* ii, P1=∑ pi, i=0,1 ..., Th；UTemp2=∑ px*j, P2=∑ pj, j=Th+1 ..., 255；
Whole image averaging gray scale U：U=P1*UTemp1+P2*UTemp2；Region F1, F2Average gray U1, U2Respectively：U1 =UTemp1/P1；U2=UTemp2/P2；Inter-class variance Di：Di=P1*(U-U1)2+P2*(U2-U)2, i=Th；Side between maximum kind Difference：D=max { Di, i=0,1 ..., 255；
Gray value when obtaining maximum between-cluster variance is defined as high threshold, gray value half is defined as Low threshold；By Successive ignition, optimal threshold value is obtained, and then determine to improve the dynamic threshold of Canny operators.
In an embodiment of the present invention, the definition threshold value Th values are 0~255.
In an embodiment of the present invention, in the step S6, contour area Threshold segmentation extracting method is：
After the marginal information for detecting image, contours extract is carried out to the edge image of gained, usable floor area is as threshold Value, chickenshit in figure after rim detection and isolated noise spot are filtered out, keep the profile of solar panel to believe Breath；Calculate the area of whole profile or partial contour, the closure that wherein cartographic represenation of area outline portion and starting point line are formed Point, it is the region gross area surrounded by the string of profile camber line and connection two-end-point that partial contour area, which is,；The calculating of area by Green theorem：I.e. Closed domain is surrounded by piecewise smooth curve L, and single order continuously can be inclined on D by function P (x, y) and Q (x, y) Lead, then have：
Wherein L is the D boundary curve for taking forward direction；With the area of green theorem zoning, if region D boundary curve For L, then：The area for the edge image that can be tried to achieve, using the Threshold segmentation of binaryzation, according to Gray value height correlation between adjacent pixel inside target or background, and the pixel in intersection both sides is on gray value There is very big difference, the setting of gray threshold is carried out to the target area with unimodal intensity profile and background area, according to setting The threshold value put, is screened to contour area, further identifies profile.
Compared to prior art, the invention has the advantages that：The present invention is improved by being pre-processed to collection figure Canny rim detections and contour area threshold process carry out profile drafting, can improve the recognition accuracy of solar panel.
Brief description of the drawings
Fig. 1 is the FB(flow block) of the present invention.
Fig. 2 is to be illustrated using the transition diagram picture of each step of the inventive method；Wherein：Fig. 2 a) it is that solar cell interpolation is taken out Image after sample；Fig. 2 b) be after filtering with the image after the pretreatment operation such as color space conversion；Fig. 2 c) it is by unrelated The image that point excludes；Fig. 2 d) it is image of the solar panel after morphology opens operation；Fig. 2 e) it is to be calculated by improving Canny The image of sub- rim detection；Fig. 2 f) it is image by contour area threshold process；Fig. 2 g) it is that solar panel profile is known Other image.
Embodiment
Below in conjunction with the accompanying drawings, technical scheme is specifically described.
The present invention's is a kind of based on the solar panel recognition methods for improving Canny operators and contour area threshold value, bag Include following steps,
Step S1, the image of solar panel is gathered by image capture device, is stored in computer, and using slotting The method of value sampling is adjusted to the size and pixel of image；
Step S2, sampled via the interpolation in step S1, be filtered operation to image by the way of gaussian filtering；
Step S3, wrapped via the filtering operation in step S2, the color space of transition diagram picture in follow-up operation Include brightness, the colouring information of saturation degree；
Step S4, to step S3 converting colors space afterwards, it is necessary to be excluded to Independent Point；After exclusion Independent Point Figure carry out morphology and open operation, holding overall intensity level and larger bright features are relatively constant；
Step S5, the image obtained to step S4 processing, rim detection is carried out to image using improved Canny operators and obtained To edge image；
Step S6, the edge image obtained to step S5 processing, the wheel of its image is extracted using contour area Threshold segmentation Wide information, the contour line of solar panel is drawn out, complete the identifying purpose of solar panel.
Gaussian filtering method in the step S2 is：The profile information of image redundancy is entered by the way of gaussian filtering Row is filtered out, and during gaussian filtering, each pixel is obtained after being weighted averagely by other pixels in itself and neighborhood Arrive, i.e. the Gaussian Blur process of image is exactly that image does convolution with normal distribution.
In the step S5, rim detection is carried out to image using improved Canny operators and obtains the method for edge image For：
By improving the objective selected threshold of Canny operators, to cause Threshold segmentation to divide target area and background area, make The dynamic threshold that the noise spot of erroneous segmentation is as few as possible, is determined using maximum between-cluster variance must occur：
The probability shared by each gray value of image is determined first, and the probability for defining gray scale i is pi, define threshold value Th (definition Threshold value Th values are 0~255), travel through Th=0, Th=1 ... ..., Th=255；
Image is divided into by two parts according to threshold value Th, is defined as F1, F2；Wherein：F1=f (x, y) | f (x, y) >=Th }；F2 =f (x, y) | f (x, y) ＜ Th }；
F is calculated respectively1, F2The average value UTemp of respective gray scale1, UTemp2, respective shared probability P1, P2；It is defined as：
UTemp1=∑ pi* ii, P1=∑ pi, i=0,1 ..., Th；UTemp2=∑ pj* J, P2=∑ pj, j=Th+ 1 ..., 255；
Whole image averaging gray scale U：U=P1*UTemp1+P2*UTemp2；Region F1, F2Average gray U1, U2Respectively：U1 =UTemp1/P1；U2=UTemp2/P2；Inter-class variance Di：Di=P1*(U-U1)2+P2*(U2-U)2, i=Th；Side between maximum kind Difference：D=max { Di, i=0,1 ..., 255；
Gray value when obtaining maximum between-cluster variance is defined as high threshold, gray value half is defined as Low threshold；By Successive ignition, optimal threshold value is obtained, and then determine to improve the dynamic threshold of Canny operators.
In the step S6, contour area Threshold segmentation extracting method is：
After the marginal information for detecting image, contours extract is carried out to the edge image of gained, usable floor area is as threshold Value, chickenshit in figure after rim detection and isolated noise spot are filtered out, keep the profile of solar panel to believe Breath；Calculate the area of whole profile or partial contour, the closure that wherein cartographic represenation of area outline portion and starting point line are formed Point, it is the region gross area surrounded by the string of profile camber line and connection two-end-point that partial contour area, which is,；The calculating of area by Green theorem：I.e. Closed domain is surrounded by piecewise smooth curve L, and single order continuously can be inclined on D by function P (x, y) and Q (x, y) Lead, then have：
Wherein L is the D boundary curve for taking forward direction；With the area of green theorem zoning, if region D boundary curve For L, then：The area for the edge image that can be tried to achieve, using the Threshold segmentation of binaryzation, according to Gray value height correlation between adjacent pixel inside target or background, and the pixel in intersection both sides is on gray value There is very big difference, the setting of gray threshold is carried out to the target area with unimodal intensity profile and background area, according to setting The threshold value put, is screened to contour area, further identifies profile.
It is below the specific embodiment of the present invention.
Present embodiments provide a kind of based on the solar panel identification for improving Canny operators and contour area threshold process Method, FB(flow block) are as shown in Figure 1.This method to collection picture by entering row interpolation sampling, denoising and converting colors space Etc. pretreatment operation, Independent Point exclusion, improved Canny rim detections and contour area threshold process is utilized to carry out solar-electricity Pond face plate edge extraction, identifies solar panel.Specifically include following steps：
Step S1：The picture of solar panel is gathered by image capture device, is stored in computer, due to storage The definition of picture is high, and shared memory space is larger, and the method sampled using interpolation is adjusted to the size and pixel of image It is whole；
Step S2：Sampled via the interpolation in step S1, still there is more profile information in picture, filtered using Gauss The mode of ripple is filtered operation to image；
Step S3：Via the filtering operation in step S2, the color space of translated image can obtain in follow-up operation To colouring informations such as brightness, saturation degrees；
Step S4：After changing color space to step S3, the background area of image, which has some colouring informations, to disturb Rim detection, so needing to exclude Independent Point.Exclude Independent Point after figure carry out morphology open operation, remove compared with Small bright detail, keep overall intensity level and larger bright features relatively constant；
Step S5：The image obtained to step S4 processing, rim detection is carried out to image using improved Canny operators；
Step S6：The edge image obtained to step S5 processing, the wheel of its image is extracted using contour area Threshold segmentation Wide information, draws out the contour line of solar panel, and the identifying purpose of this solar panel is reached.
Fig. 2 a) for the present embodiment sampled as interpolation after obtained by solar panel image.Due to smart mobile phone or Picture clarity captured by other high-definition cameras of person is very high, and profile details are more, and shared memory space is larger, is unfavorable for too The identification of positive energy cell panel target area, the method sampled using interpolation are adjusted to the size and pixel of image, interpolation side The efficiency that bicubic interpolation is realized in formula is although relatively low, but effect is fine, so using bicubic interpolation.Bicubic interpolation is protected Preferable image detail has been stayed, the purpose for taking into account efficiency and effect can be reached.
Image in the present embodiment after interpolation sampling needs to carry out pretreatment operation, mainly includes denoising and color space Conversion.The profile information of redundancy still compares more in image, is filtered operation, Gauss to image by the way of gaussian filtering In filtering, each pixel is obtained after being weighted averagely by other pixels in itself and neighborhood, i.e. image Gaussian Blur process be exactly image and convolution is done in normal distribution.HSV is a kind of cone-shaped model, and the model is by form and aspect, saturation degree Described with brightness.Due to HSV acted on when carrying out formulation color segmentation it is larger, so RGB is converted into HSV moulds by the present embodiment Type.Fig. 2 b) it is the image obtained after step 3 pretreatment operation.
In the present embodiment there is the meeting Clutter edge detection of some colouring informations, it is necessary to unrelated click-through in the background area of image Row excludes, and strengthens target area, weakens background area.Fig. 2 c) it is the image obtained after step 4 exclusive PCR point.
Morphology, which opens operation, can remove less bright detail, keep overall intensity level and larger bright features relative Constant, its formula is：The formula, which represents first only to be F of B, to be corroded, then with B to gained As a result do and expand.Carried out so the present embodiment opens operation using morphology, exemplary plot such as Fig. 2 d) shown in.
The improvement Canny operators, the step of based on Canny rim detections：Gaussian smoothing removes noise, calculates gradient width Value and direction, non-maxima suppression, hysteresis threshold.The key of the present invention is objective selected threshold.Threshold segmentation should be fine Ground divides target area and background area so that it is as few as possible the noise spot of erroneous segmentation occur, true using maximum between-cluster variance Fixed dynamic threshold is higher compared to traditional Canny operator repetition test threshold efficiencies.The each gray value institute of image is determined first The probability accounted for, the probability for defining gray scale i is Pi, define threshold value Th (wherein Th values are 0~255).Travel through Th=0, Th= 1 ... ..., Th=255.Image is divided into by two parts according to threshold value Th, is defined as F1、F2.Wherein：F1=f (x, y) | f (x, y) ≥Th}；F2=f (x, y) | f (x, y) ＜ Th }.F is calculated respectively1, F2The average value UTemp of respective gray scale1, UTemp2, respective institute The probability P accounted for1, P2.It is defined as：UTemp1=∑ pi* i, P1=∑ pi, i=0,1 ..., Th；UTemp2=∑ pj* j, P2=∑ pj, j=Th+1 ..., 255.
Whole image averaging gray scale U：U=P1*UTemp1+P2*UTemp2.Region F1, F2Average gray U1, U2Respectively：
U1=UTemp1/P1；U2=UTemp2/P2.Inter-class variance Di：Di=P1*(U-U1)2+P2*(U2-U)2, i=Th.Most Big inter-class variance：D=max { Di, i=0, gray value when obtaining maximum between-cluster variance is defined as high threshold by 1 ..., 255., will Gray value half is defined as Low threshold.Need by successive ignition, can just obtain optimal threshold value, and then determine to improve Canny The dynamic threshold of operator.Improve Canny operators and carry out rim detection such as Fig. 2 e) shown in.
After the marginal information for detecting image, contours extract is carried out to the image of gained, usable floor area, will as threshold value Chickenshit and isolated noise spot filter out in figure after rim detection, keep the preferable profile letter of solar panel Breath.Calculate the area of whole profile or partial contour, the closure that wherein cartographic represenation of area outline portion and starting point line are formed Point, it is the region gross area surrounded by the string of profile camber line and connection two-end-point that partial contour area, which is,.The calculating of area by Green theorem：I.e. Closed domain is surrounded by piecewise smooth curve L, and single order continuously can be inclined on D by function P (x, y) and Q (x, y) Lead, then have：
Wherein L is the D boundary curve for taking forward direction.With the area of green theorem zoning, if region D boundary curve For L, then：
The area for the edge image that can be tried to achieve, using the Threshold segmentation of binaryzation, according to inside target or background Gray value height correlation between adjacent pixel, and the pixel in intersection both sides has very big difference on gray value, to tool There are the target area of unimodal intensity profile and background area to carry out the setting of gray threshold, according to the threshold value of setting, to contoured surface Product is screened, and further identifies profile.Exemplary plot such as Fig. 2 g) shown in.
The basic procedure and important step of present invention described above, only presently preferred embodiments of the present invention and exemplary retouch Stating, the technical staff in photovoltaic image technique field can carry out some improvement and equivalence replacement after present patent application is read, Protection scope of the present invention should be included to the various modifications of present invention progress by not departing from the scope that the present invention relates to.

Claims (5)

  1. It is 1. a kind of based on the solar panel recognition methods for improving Canny operators and contour area threshold value, it is characterised in that：Bag Include following steps,
    Step S1, the image of solar panel is gathered by image capture device, is stored in computer, and is taken out using interpolation The method of sample is adjusted to the size and pixel of image；
    Step S2, sampled via the interpolation in step S1, be filtered operation to image by the way of gaussian filtering；
    Step S3, via the filtering operation in step S2, the color space of transition diagram picture obtains including bright in follow-up operation Degree, the colouring information of saturation degree；
    Step S4, to step S3 converting colors space afterwards, it is necessary to be excluded to Independent Point；To the figure after exclusion Independent Point Shape carries out morphology and opens operation, keeps overall intensity level and larger bright features relatively constant；
    Step S5, the image obtained to step S4 processing, rim detection is carried out to image using improved Canny operators and obtains side Edge image；
    Step S6, the edge image obtained to step S5 processing, the profile that its image is extracted using contour area Threshold segmentation are believed Breath, the contour line of solar panel is drawn out, complete the identifying purpose of solar panel.
  2. It is 2. according to claim 1 based on the solar panel identification side for improving Canny operators and contour area threshold value Method, it is characterised in that：Gaussian filtering method in the step S2 is：To the profile of image redundancy by the way of gaussian filtering Information is filtered out, and during gaussian filtering, each pixel is to be weighted putting down by other pixels in itself and neighborhood Obtain afterwards, i.e. the Gaussian Blur process of image is exactly that image does convolution with normal distribution.
  3. It is 3. according to claim 1 based on the solar panel identification side for improving Canny operators and contour area threshold value Method, it is characterised in that：In the step S5, rim detection is carried out to image using improved Canny operators and obtains edge image Method be：
    By improving the objective selected threshold of Canny operators, to cause Threshold segmentation to divide target area and background area so that go out The noise spot of existing erroneous segmentation is as few as possible, the dynamic threshold determined using maximum between-cluster variance：
    The probability shared by each gray value of image is determined first, and the probability for defining gray scale i is pi, threshold value Th is defined, travels through Th=0, Th=1 ... ..., Th=255；
    Image is divided into by two parts according to threshold value Th, is defined as F1, F1；Wherein：F1=f (x, y) | f (x, y) >=Th }；F2={ f (x, y) | f (x, y) ＜ Th }；
    F is calculated respectively1, F2The average value UTemp of respective gray scale1, UTemp2, respective shared probability P1, P2；It is defined as：UTemp1 =∑ pi* ii, P1=∑ pi, i=0,1 ..., Th；UTemp2=∑ pj* j, P2=∑ pj, j=Th+1 ..., 255；
    Whole image averaging gray scale U：U=P1*UTemp1+P2*UTemp2；Region F1, F2Average gray U1, U2Respectively：U1= UTemp1/P1；U2=UTemp2/P2；Inter-class variance Di：Di=P1*(U-U1)2+P2*(U2-U)2, i=Th；Maximum between-cluster variance：D =max { Di, i=0,1 ..., 255；
    Gray value when obtaining maximum between-cluster variance is defined as high threshold, gray value half is defined as Low threshold；By multiple Iteration, optimal threshold value is obtained, and then determine to improve the dynamic threshold of Canny operators.
  4. It is 4. according to claim 3 based on the solar panel identification side for improving Canny operators and contour area threshold value Method, it is characterised in that：The definition threshold value Th values are 0~255.
  5. It is 5. according to claim 1 based on the solar panel identification side for improving Canny operators and contour area threshold value Method, it is characterised in that：In the step S6, contour area Threshold segmentation extracting method is：
    After the marginal information for detecting image, contours extract is carried out to the edge image of gained, usable floor area, will as threshold value Chickenshit and isolated noise spot filter out in figure after rim detection, keep the profile information of solar panel；Meter Calculate the area of whole profile or partial contour, the enclosure portion that wherein cartographic represenation of area outline portion and starting point line are formed, portion It is the region gross area to be surrounded by the string of profile camber line and connection two-end-point to divide contour area；The calculating of area is public by Green Formula：I.e. Closed domain is surrounded by piecewise smooth curve L, function P (x, y) and Q (x, y) on D single order continuously can local derviation, then have：
    Wherein L is the D boundary curve for taking forward direction；With the area of green theorem zoning, if region D boundary curve is L, Then：The area for the edge image that can be tried to achieve, using the Threshold segmentation of binaryzation, according in Gray value height correlation inside target or background between adjacent pixel, and the pixel in intersection both sides has very on gray value Big difference, the setting of gray threshold is carried out to the target area with unimodal intensity profile and background area, according to setting Threshold value, contour area is screened, further identify profile.
CN201711313921.3A 2017-12-11 2017-12-11 Based on the solar panel recognition methods for improving Canny operators and contour area threshold value Pending CN107895376A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711313921.3A CN107895376A (en) 2017-12-11 2017-12-11 Based on the solar panel recognition methods for improving Canny operators and contour area threshold value

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711313921.3A CN107895376A (en) 2017-12-11 2017-12-11 Based on the solar panel recognition methods for improving Canny operators and contour area threshold value

Publications (1)

Publication Number Publication Date
CN107895376A true CN107895376A (en) 2018-04-10

Family

ID=61806231

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711313921.3A Pending CN107895376A (en) 2017-12-11 2017-12-11 Based on the solar panel recognition methods for improving Canny operators and contour area threshold value

Country Status (1)

Country Link
CN (1) CN107895376A (en)

Cited By (29)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108550113A (en) * 2018-04-17 2018-09-18 东莞市金翔光电科技有限公司 Image scanning output method, device, computer equipment and storage medium
CN108830874A (en) * 2018-04-19 2018-11-16 麦克奥迪（厦门）医疗诊断系统有限公司 A kind of number pathology full slice Image blank region automatic division method
CN108996268A (en) * 2018-08-01 2018-12-14 上海主线科技有限公司 A kind of container tractor based on camera and suspension bridge are mutually located method
CN109636824A (en) * 2018-12-20 2019-04-16 巢湖学院 A kind of multiple target method of counting based on image recognition technology
CN109829919A (en) * 2019-01-31 2019-05-31 苏州晟成光伏设备有限公司 A kind of vision positioning method of solar battery sheet
CN110391779A (en) * 2019-05-04 2019-10-29 刘纪君 The controllable type angle modification system of solar street light
CN110766675A (en) * 2019-10-22 2020-02-07 孟帅帅 Solar cell panel defect detection method
CN110910316A (en) * 2019-05-14 2020-03-24 程爱军 Solar equipment control platform
CN110991458A (en) * 2019-11-25 2020-04-10 创新奇智(北京)科技有限公司 Artificial intelligence recognition result sampling system and sampling method based on image characteristics
CN111027567A (en) * 2019-10-30 2020-04-17 四川轻化工大学 Edge extraction method based on algorithm learning
CN111179289A (en) * 2019-12-31 2020-05-19 重庆邮电大学 Image segmentation method suitable for webpage length and width images
CN111209842A (en) * 2020-01-02 2020-05-29 珠海格力电器股份有限公司 Visual positioning processing method and device and robot
CN111311610A (en) * 2020-02-14 2020-06-19 河北工程大学 Image segmentation method and terminal equipment
CN111476792A (en) * 2020-05-27 2020-07-31 东北大学 Method for extracting plate strip steel image outline
CN111523391A (en) * 2020-03-26 2020-08-11 上海刻羽信息科技有限公司 Building identification method, system, electronic device and readable storage medium
CN112053322A (en) * 2020-07-31 2020-12-08 上海电机学院 Method for segmenting and detecting surface shielding of photovoltaic cell panel
CN112150492A (en) * 2019-06-26 2020-12-29 司空定制家居科技有限公司 Method and device for reading house-type graph and storage medium
CN112258548A (en) * 2020-10-20 2021-01-22 东南大学 Moving object extraction method based on improved ViBe algorithm
CN112330643A (en) * 2020-11-10 2021-02-05 国网湖北省电力有限公司宜昌供电公司 Secondary equipment state identification method based on sparse representation image restoration
CN112381084A (en) * 2020-10-12 2021-02-19 武汉沃亿生物有限公司 Automatic contour recognition method for tomographic image
CN112648920A (en) * 2019-10-12 2021-04-13 上海微电子装备（集团）股份有限公司 Mask opening size measuring method, mask plate stretching device and screen expanding machine
CN112669295A (en) * 2020-12-30 2021-04-16 上海电机学院 Lithium battery pole piece defect detection method based on secondary threshold segmentation theory
CN113034529A (en) * 2021-04-02 2021-06-25 广州绿怡信息科技有限公司 Equipment image extraction method and device based on mini-mobile phone detector
CN113361382A (en) * 2021-05-14 2021-09-07 沈阳工业大学 Hand shape recognition method based on compressed relative contour feature points
CN113610799A (en) * 2021-08-04 2021-11-05 沭阳九鼎钢铁有限公司 Artificial intelligence-based photovoltaic cell panel rainbow line detection method, device and equipment
CN113807293A (en) * 2021-09-24 2021-12-17 纵目科技(重庆)有限公司 Deceleration strip detection method, system, equipment and computer readable storage medium
CN114820475A (en) * 2022-04-11 2022-07-29 苏州优力科瑞半导体科技有限公司 Edge identification method and system, wafer processing device and method for determining concentric state of wafer and processing table
CN115063421A (en) * 2022-08-16 2022-09-16 成都数联云算科技有限公司 Pole piece region detection method, system and device, medium and defect detection method
CN115330768A (en) * 2022-10-12 2022-11-11 江苏跃格智能装备有限公司 Quality grading method for solar cell panel

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102974551A (en) * 2012-11-26 2013-03-20 华南理工大学 Machine vision-based method for detecting and sorting polycrystalline silicon solar energy
CN103872983A (en) * 2014-04-04 2014-06-18 天津市鑫鼎源科技发展有限公司 Device and method for detecting defects on surface of solar cell
CN104166841A (en) * 2014-07-24 2014-11-26 浙江大学 Rapid detection identification method for specified pedestrian or vehicle in video monitoring network
CN107014819A (en) * 2017-06-09 2017-08-04 杭州电子科技大学 A kind of solar panel surface defects detection system and method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102974551A (en) * 2012-11-26 2013-03-20 华南理工大学 Machine vision-based method for detecting and sorting polycrystalline silicon solar energy
CN103872983A (en) * 2014-04-04 2014-06-18 天津市鑫鼎源科技发展有限公司 Device and method for detecting defects on surface of solar cell
CN104166841A (en) * 2014-07-24 2014-11-26 浙江大学 Rapid detection identification method for specified pedestrian or vehicle in video monitoring network
CN107014819A (en) * 2017-06-09 2017-08-04 杭州电子科技大学 A kind of solar panel surface defects detection system and method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
FENGJIE WU ET AL.: ""Aerial image recognition and matching for inspection of large-scale photovoltaic farms"", 《IEEE》 *
唐路路 等: ""一种自适应阈值的 Canny 边缘检测算法"", 《光电工程》 *
胡佳成 等: ""ABS 齿圈环形表面缺陷检测方法"", 《电子测量与仪器学报》 *
陈漱文: ""基于图像处理的一种灯阵排布识别方法"", 《万方数据知识服务平台》 *

Cited By (43)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108550113A (en) * 2018-04-17 2018-09-18 东莞市金翔光电科技有限公司 Image scanning output method, device, computer equipment and storage medium
CN108830874A (en) * 2018-04-19 2018-11-16 麦克奥迪（厦门）医疗诊断系统有限公司 A kind of number pathology full slice Image blank region automatic division method
CN108996268A (en) * 2018-08-01 2018-12-14 上海主线科技有限公司 A kind of container tractor based on camera and suspension bridge are mutually located method
CN109636824B (en) * 2018-12-20 2022-10-11 巢湖学院 Multi-target counting method based on image recognition technology
CN109636824A (en) * 2018-12-20 2019-04-16 巢湖学院 A kind of multiple target method of counting based on image recognition technology
CN109829919A (en) * 2019-01-31 2019-05-31 苏州晟成光伏设备有限公司 A kind of vision positioning method of solar battery sheet
CN110391779B (en) * 2019-05-04 2021-05-11 扬州市文灏光伏科技有限公司 Controllable angle correction system of solar street lamp
CN110391779A (en) * 2019-05-04 2019-10-29 刘纪君 The controllable type angle modification system of solar street light
CN110910316A (en) * 2019-05-14 2020-03-24 程爱军 Solar equipment control platform
CN112150492A (en) * 2019-06-26 2020-12-29 司空定制家居科技有限公司 Method and device for reading house-type graph and storage medium
CN112648920A (en) * 2019-10-12 2021-04-13 上海微电子装备（集团）股份有限公司 Mask opening size measuring method, mask plate stretching device and screen expanding machine
CN110766675A (en) * 2019-10-22 2020-02-07 孟帅帅 Solar cell panel defect detection method
CN110766675B (en) * 2019-10-22 2020-07-10 科士恩科技(上海)有限公司 Solar cell panel defect detection method
CN111027567A (en) * 2019-10-30 2020-04-17 四川轻化工大学 Edge extraction method based on algorithm learning
CN110991458A (en) * 2019-11-25 2020-04-10 创新奇智(北京)科技有限公司 Artificial intelligence recognition result sampling system and sampling method based on image characteristics
CN111179289A (en) * 2019-12-31 2020-05-19 重庆邮电大学 Image segmentation method suitable for webpage length and width images
CN111179289B (en) * 2019-12-31 2023-05-19 重庆邮电大学 Image segmentation method suitable for webpage length graph and width graph
CN111209842B (en) * 2020-01-02 2023-06-30 珠海格力电器股份有限公司 Visual positioning processing method and device and robot
CN111209842A (en) * 2020-01-02 2020-05-29 珠海格力电器股份有限公司 Visual positioning processing method and device and robot
CN111311610A (en) * 2020-02-14 2020-06-19 河北工程大学 Image segmentation method and terminal equipment
CN111523391B (en) * 2020-03-26 2021-11-05 上海刻羽信息科技有限公司 Building identification method, system, electronic device and readable storage medium
CN111523391A (en) * 2020-03-26 2020-08-11 上海刻羽信息科技有限公司 Building identification method, system, electronic device and readable storage medium
CN111476792B (en) * 2020-05-27 2023-05-23 东北大学 Extraction method of strip steel image contour
CN111476792A (en) * 2020-05-27 2020-07-31 东北大学 Method for extracting plate strip steel image outline
CN112053322A (en) * 2020-07-31 2020-12-08 上海电机学院 Method for segmenting and detecting surface shielding of photovoltaic cell panel
CN112053322B (en) * 2020-07-31 2022-07-22 上海电机学院 Method for segmenting and detecting surface shielding of photovoltaic cell panel
CN112381084A (en) * 2020-10-12 2021-02-19 武汉沃亿生物有限公司 Automatic contour recognition method for tomographic image
CN112381084B (en) * 2020-10-12 2024-02-09 武汉沃亿生物有限公司 Automatic contour recognition method for tomographic image
CN112258548B (en) * 2020-10-20 2024-03-29 东南大学 Moving target extraction method based on improved ViBe algorithm
CN112258548A (en) * 2020-10-20 2021-01-22 东南大学 Moving object extraction method based on improved ViBe algorithm
CN112330643B (en) * 2020-11-10 2023-02-07 国网湖北省电力有限公司宜昌供电公司 Secondary equipment state identification method based on sparse representation image restoration
CN112330643A (en) * 2020-11-10 2021-02-05 国网湖北省电力有限公司宜昌供电公司 Secondary equipment state identification method based on sparse representation image restoration
CN112669295A (en) * 2020-12-30 2021-04-16 上海电机学院 Lithium battery pole piece defect detection method based on secondary threshold segmentation theory
CN113034529A (en) * 2021-04-02 2021-06-25 广州绿怡信息科技有限公司 Equipment image extraction method and device based on mini-mobile phone detector
CN113361382B (en) * 2021-05-14 2024-02-02 沈阳工业大学 Hand shape recognition method based on compressed relative contour feature points
CN113361382A (en) * 2021-05-14 2021-09-07 沈阳工业大学 Hand shape recognition method based on compressed relative contour feature points
CN113610799A (en) * 2021-08-04 2021-11-05 沭阳九鼎钢铁有限公司 Artificial intelligence-based photovoltaic cell panel rainbow line detection method, device and equipment
CN113807293B (en) * 2021-09-24 2024-02-09 纵目科技(重庆)有限公司 Deceleration strip detection method, deceleration strip detection system, deceleration strip detection equipment and computer readable storage medium
CN113807293A (en) * 2021-09-24 2021-12-17 纵目科技(重庆)有限公司 Deceleration strip detection method, system, equipment and computer readable storage medium
CN114820475A (en) * 2022-04-11 2022-07-29 苏州优力科瑞半导体科技有限公司 Edge identification method and system, wafer processing device and method for determining concentric state of wafer and processing table
CN115063421B (en) * 2022-08-16 2022-10-28 成都数联云算科技有限公司 Pole piece region detection method, system and device, medium and defect detection method
CN115063421A (en) * 2022-08-16 2022-09-16 成都数联云算科技有限公司 Pole piece region detection method, system and device, medium and defect detection method
CN115330768A (en) * 2022-10-12 2022-11-11 江苏跃格智能装备有限公司 Quality grading method for solar cell panel

Similar Documents

Publication Publication Date Title
CN107895376A (en) Based on the solar panel recognition methods for improving Canny operators and contour area threshold value
CN109558806B (en) Method for detecting high-resolution remote sensing image change
WO2018107939A1 (en) Edge completeness-based optimal identification method for image segmentation
Zhang et al. Object-oriented shadow detection and removal from urban high-resolution remote sensing images
Chen et al. A novel color edge detection algorithm in RGB color space
CN105184779B (en) One kind is based on the pyramidal vehicle multiscale tracing method of swift nature
CN107578035A (en) Human body contour outline extracting method based on super-pixel polychrome color space
WO2020007307A1 (en) Sky filter method for panoramic images and portable terminal
CN102974551A (en) Machine vision-based method for detecting and sorting polycrystalline silicon solar energy
CN107507208A (en) A kind of characteristics of image point extracting method based on Curvature Estimation on profile
CN107944354B (en) Vehicle detection method based on deep learning
CN107185854A (en) The algorithm of photovoltaic cell acetes chinensis and color classification based on RGB channel
CN106294705A (en) A kind of batch remote sensing image preprocess method
CN102609917A (en) Image edge fitting B spline generating method based on clustering algorithm
CN105719275A (en) Parallel combination image defect segmentation method
CN113537211B (en) Asymmetric IOU-based deep learning license plate frame positioning method
CN109801283A (en) A kind of Hydrophobicity of Composite Insulator grade determination method based on water droplet offset distance
CN107273884A (en) A kind of License Plate Identification method based on mobile terminal camera
CN104992448B (en) The automatic positioning method of robot antisitic defect grape-picking
CN112861654A (en) Famous tea picking point position information acquisition method based on machine vision
CN112435272A (en) High-voltage transmission line connected domain removing method based on image contour analysis
CN108492306A (en) A kind of X-type Angular Point Extracting Method based on image outline
CN112232249B (en) Remote sensing image change detection method and device based on depth characteristics
CN105787955A (en) Sparse segmentation method and device of strip steel defect
CN105894501B (en) A kind of detection of high-resolution remote sensing image list wood and tree crown plotting method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20180410